This paper describes the use of a scanning electron microscope (SEM)/X-ray/image-analysis system to perform automatic petrographic analyses of sandstone samples. The methods developed involve the combined use of backscattered-electron (BSE) imaging and X-ray analysis with image-analysis procedures to quantify the mineralogy, grain texture, and pore structure of polished rock samples. The procedures are fully automated and, once set up, can be run without operator involvement (e.g., overnight). A wealth of quantitative data can be obtained with this single approach, providing useful information for many fields of petroleum engineering (diagenetic studies, prediction of physical properties, etc.). These methods should help replace, at least partially, the tedious and timeconsuming manual petrographic methods currently used.
IntroductionPetrographic analysis of sedimentary rocks routinely involves the use of an optical microscope to examine thin-sections. Often, an SEM is also used to obtain more detailed information from broken rock surfaces and BSE imaging of resin-impregnated and polished samples. 1,2 With one or a combination of these techniques, qualitative descriptions of mineralogy, grain texture (size and shape), and pore structure can be obtained. Such studies provide information about controls on reservoir quality3,4 and rock/fluid interaction during drilling and production operations. 5 Quantitative or semiquantitative petrographic information requires more tedious and time-consuming manual procedures-e.g., point counting to determine mineralogy.To reduce the labor involved in such techniques, automatic rock characterization with an SEM/X-ray/image-analysis system was investigated. The results can be used in both petrophysical and reservoir geological models to derive reservoir parameters (e.g., permeability) and to investigate their relationships to geological variables.Previously, Dilks and Graham 6 used BSE imaging to help quantify the mineralogy of sandstone samples. The use of this technique alone, however, does not allow for complete discrimination of every mineral. To overcome this, Minnis 7 used X -ray analysis to determine chemistry, and hence mineralogy, but because the mineral analyses were done over an array of widely spaced points, only limited spatial and textural information could be obtained. Petruk 8 extended this work by using both BSE imaging and X-ray analysis to recognize ore minerals. Image analysis has also been used to characterize pore structure.